Evaluation of Nystagmus Examination Using Wearable AR Glasses in Vertigo Patients

Overview

Background and Purpose: Vertigo is common in emergency and outpatient settings, yet standard oculomotor testing usually requires dedicated equipment and exam rooms. This study evaluates whether nystagmus examinations performed with wearable augmented-reality (AR) glasses are equivalent to conventional examination-room testing for classifying central vs. peripheral vertigo. The investigators also assess diagnostic accuracy, patient tolerability, and the reliability of AR-based interpretation (test-retest and inter-rater). Study Design: Prospective, single-center, within-subject randomized equivalence study at Kaohsiung Chang Gung Memorial Hospital. Each participant completes both AR-based and conventional oculomotor testing in a randomized order during the same visit, separated by a 30-minute washout. The study uses an evaluator-blind approach: de-identified trajectories are reviewed offline by independent experts who are masked to test modality. A follow-up visit (\~1 week) captures adverse events and patient experience. Participants: Adults (≥18 years) presenting with vertigo who can tolerate the AR headset and provide consent. Key exclusions include conditions that prevent reliable eye-tracking (e.g., corrected visual acuity \<20/40), recent use of vestibular suppressants (within 24 hours), and other factors limiting cooperation or safety. Interventions and Procedures: The AR system records eye movements and presents standardized visual stimuli. Conventional testing follows current clinical standards (e.g., Frenzel/oculomotor exam). All recordings are stored securely for blinded review. Outcomes: Primary endpoint: Equivalence of diagnostic agreement (central vs. peripheral) between AR-based and conventional methods, quantified by Cohen's kappa (κ) with a predefined equivalence margin. Secondary endpoints: (1) Diagnostic accuracy of the AR method in a clinically-indicated imaging subgroup (MRI preferred; CT as needed) using a sequential evaluation strategy; (2) Patient discomfort/tolerability using VAS and CSQ-VR, compared between modalities; (3) Test-retest reliability of AR-based classifications; (4) Inter-rater reliability between independent evaluators, with a third reader adjudicating discordant cases; (5) prespecified subgroup analyses by age, medical history, and vestibular function. Sample Size and Duration: Approximately 200 participants will be enrolled (target \~180 evaluable after \~10% attrition). Total study duration is \~2 years, including enrollment, follow-up, and analysis. Risks and Benefits: Both tests are non-invasive. Potential transient discomfort (e.g., eye strain or cybersickness) will be monitored. There may be no direct benefit to participants; however, results could support broader, more accessible, and standardized vertigo assessment. Data Security and Privacy: All data are de-identified, stored on secure platforms with role-based access and audit trails. Safety events are monitored and graded, and protocol deviations are handled per Good Clinical Practice.